ICCV 2017 论文列表

The conventional methods for estimating camera poses and scene structures from severely blurry or low resolution images often result in failure. The off-the-shelf deblurring or super resolution methods may show visually pleasing results. However, applying each technique independently before matching is generally unprofitable because this naive series of procedures ignores the consistency between images. In this paper, we propose a pioneering unified framework that solves four problems simultaneously, namely, dense depth reconstruction, camera pose estimation, super resolution, and deblurring. By reflecting a physical imaging process, we formulate a cost minimization problem and solve it using an alternating optimization technique. The experimental results on both synthetic and real videos show high-quality depth maps derived from severely degraded images that contrast the failures of naive multi-view stereo methods. Our proposed method also produces outstanding deblurred and super-resolved images unlike the independent application or combination of conventional video deblurring, super resolution methods.

Many human activities involve object manipulations aiming to modify the object state. Examples of common state changes include full/empty bottle, open/closed door, and attached/detached car wheel. In this work, we seek to automatically discover the states of objects and the associated manipulation actions. Given a set of videos for a particular task, we propose a joint model that learns to identify object states and to localize state-modifying actions. Our model is formulated as a discriminative clustering cost with constraints. We assume a consistent temporal order for the changes in object states and manipulation actions, and introduce new optimization techniques to learn model parameters without additional supervision. We demonstrate successful discovery of seven manipulation actions and corresponding object states on a new dataset of videos depicting real-life object manipulations. We show that our joint formulation results in an improvement of object state discovery by action recognition and vice versa.

This paper addresses the problem of joint detection and recounting of abnormal events in videos. Recounting of abnormal events, i.e., explaining why they are judged to be abnormal, is an unexplored but critical task in video surveillance, because it helps human observers quickly judge if they are false alarms or not. To describe the events in the human-understandable form for event recounting, learning generic knowledge about visual concepts (e.g., object and action) is crucial. Although convolutional neural networks (CNNs) have achieved promising results in learning such concepts, it remains an open question as to how to effectively use CNNs for abnormal event detection, mainly due to the environment-dependent nature of the anomaly detection. In this paper, we tackle this problem by integrating a generic CNN model and environment-dependent anomaly detectors. Our approach first learns CNN with multiple visual tasks to exploit semantic information that is useful for detecting and recounting abnormal events. By appropriately plugging the model into anomaly detectors, we can detect and recount abnormal events while taking advantage of the discriminative power of CNNs. Our approach outperforms the state-of-the-art on Avenue and UCSD Ped2 benchmarks for abnormal event detection and also produces promising results of abnormal event recounting.

Analysis sparse representation (ASR) and synthesis sparse representation (SSR) are two representative approaches for sparsity-based image modeling. An image is described mainly by the non-zero coefficients in SSR, while it is characterized by the indices of zeros in ASR. To exploit the complementary representation mechanisms of ASR and SSR, we integrate the two models and propose a joint convolutional analysis and synthesis (JCAS) sparse representation model. The convolutional implementation is adopted to more effectively exploit the image global information. In JCAS, a single image is decomposed into two layers, one is approximated by ASR to represent image large-scale structures, and the other by SSR to represent image fine-scale textures. The synthesis dictionary is adaptively learned in JCAS to describe the texture patterns for different single image layer separation tasks. We evaluate the proposed JCAS model on a variety of applications, including rain streak removal, high dynamic range image tone mapping, etc. The results show that our JCAS method outperforms state-ofthe-arts in those applications in terms of both quantitative measure and visual perception quality.

We present a novel method for removing rain streaks from a single input image by decomposing it into a rain-free background layer B and a rain-streak layer R. A joint optimization process is used that alternates between removing rain-streak details from B and removing non-streak details from R. The process is assisted by three novel image priors. Observing that rain streaks typically span a narrow range of directions, we first analyze the local gradient statistics in the rain image to identify image regions that are dominated by rain streaks. From these regions, we estimate the dominant rain streak direction and extract a collection of rain-dominated patches. Next, we define two priors on the background layer B, one based on a centralized sparse representation and another based on the estimated rain direction. A third prior is defined on the rain-streak layer R, based on similarity of patches to the extracted rain patches. Both visual and quantitative comparisons demonstrate that our method outperforms the state-of-the-art.

Recent works on adaptive sparse and low-rank signal modeling have demonstrated their usefulness, especially in image/video processing applications. While a patch-based sparse model imposes local structure, low-rankness of the grouped patches exploits non-local correlation. Applying either approach alone usually limits performance in various low-level vision tasks. In this work, we propose a novel video denoising method, based on an online tensor reconstruction scheme with a joint adaptive sparse and low-rank model, dubbed SALT. An efficient and unsupervised online unitary sparsifying transform learning method is introduced to impose adaptive sparsity on the fly. We develop an efficient 3D spatio-temporal data reconstruction framework based on the proposed online learning method, which exhibits low latency and can potentially handle streaming videos. To the best of our knowledge, this is the first work that combines adaptive sparsity and low-rankness for video denoising, and the first work of solving the proposed problem in an online fashion. We demonstrate video denoising results over commonly used videos from public datasets. Numerical experiments show that the proposed video denoising method outperforms competing methods.

By stacking layers of convolution and nonlinearity, convolutional networks (ConvNets) effectively learn from low-level to high-level features and discriminative representations. Since the end goal of large-scale recognition is to delineate complex boundaries of thousands of classes, adequate exploration of feature distributions is important for realizing full potentials of ConvNets. However, state-of-the-art works concentrate only on deeper or wider architecture design, while rarely exploring feature statistics higher than first-order. We take a step towards addressing this problem. Our method consists in covariance pooling, instead of the most commonly used first-order pooling, of high-level convolutional features. The main challenges involved are robust covariance estimation given a small sample of large-dimensional features and usage of the manifold structure of covariance matrices. To address these challenges, we present a Matrix Power Normalized Covariance (MPN-COV) method. We develop forward and backward propagation formulas regarding the nonlinear matrix functions such that MPN-COV can be trained end-to-end. In addition, we analyze both qualitatively and quantitatively its advantage over the well-known Log-Euclidean metric. On the ImageNet 2012 validation set, by combining MPN-COV we achieve over 4%, 3% and 2.5% gains for AlexNet, VGG-M and VGG-16, respectively; integration of MPN-COV into 50-layer ResNet outperforms ResNet-101 and is comparable to ResNet-152. The source code will be available on the project page: http://www.peihuali.org/MPN-COV.

We study unsupervised learning by developing a generative model built from progressively learned deep convolutional neural networks. The resulting generator is additionally a discriminator, capable of "introspection" in a sense --- being able to self-evaluate the difference between its generated samples and the given training data. Through repeated discriminative learning, desirable properties of modern discriminative classifiers are directly inherited by the generator. Specifically, our model learns a sequence of CNN classifiers using a synthesis-by-classification algorithm. In the experiments, we observe encouraging results on a number of applications including texture modeling, artistic style transferring, face modeling, and unsupervised feature learning.

We introduce a novel method to obtain high-quality 3D reconstructions from consumer RGB-D sensors. Our core idea is to simultaneously optimize for geometry encoded in a signed distance field (SDF), textures from automatically-selected keyframes, and their camera poses along with material and scene lighting. To this end, we propose a joint surface reconstruction approach that is based on Shape-from-Shading (SfS) techniques and utilizes the estimation of spatially-varying spherical harmonics (SVSH) from subvolumes of the reconstructed scene. Through extensive examples and evaluations, we demonstrate that our method dramatically increases the level of detail in the reconstructed scene geometry and contributes highly to consistent surface texture recovery.

3D dynamic surface tracking is an important research problem and plays a vital role in many computer vision and medical imaging applications. However, it is still challenging to efficiently register surface sequences which has large deformations and strong noise. In this paper, we propose a novel automatic method for non-rigid 3D dynamic surface tracking with surface Ricci flow and Teichmuller map methods. According to quasi-conformal Teichmuller theory, the Techmuller map minimizes the maximal dilation so that our method is able to automatically register surfaces with large deformations. Besides, the adoption of Delaunay triangulation and quadrilateral meshes makes our method applicable to low quality meshes. In our work, the 3D dynamic surfaces are acquired by a high speed 3D scanner. We first identified sparse surface features using machine learning methods in the texture space. Then we assign landmark features with different curvature settings and the Riemannian metric of the surface is computed by the dynamic Ricci flow method, such that all the curvatures are concentrated on the feature points and the surface is flat everywhere else. The registration among frames is computed by the Teichmuller mappings, which aligns the feature points with least angle distortions. We apply our new method to multiple sequences of 3D facial surfaces with large expression deformations and compare them with two other state-of-the-art tracking methods. The effectiveness of our method is demonstrated by the clearly improved accuracy and efficiency.

Deep feature spaces have the capacity to encode complex transformations of their input data. However, understanding the relative feature-space relationship between two transformed encoded images is difficult. For instance, what is the relative feature space relationship between two rotated images? What is decoded when we interpolate in feature space? Ideally, we want to disentangle confounding factors, such as pose, appearance, and illumination, from object identity. Disentangling these is difficult because they interact in very nonlinear ways. We propose a simple method to construct a deep feature space, with explicitly disentangled representations of several known transformations. A person or algorithm can then manipulate the disentangled representation, for example, to re-render an image with explicit control over parameterized degrees of freedom. The feature space is constructed using a transforming encoder-decoder network with a custom feature transform layer, acting on the hidden representations. We demonstrate the advantages of explicit disentangling on a variety of datasets and transformations, and as an aid for traditional tasks, such as classification.

Deep neural perception and control networks are likely to be a key component of self-driving vehicles. These models need to be explainable - they should provide easy-to-interpret rationales for their behavior - so that passengers, insurance companies, law enforcement, developers etc., can understand what triggered a particular behavior. Here we explore the use of visual explanations. These explanations take the form of real-time highlighted regions of an image that causally influence the network's output (steering control). Our approach is two-stage. In the first stage, we use a visual attention model to train a convolution network end-to-end from images to steering angle. The attention model highlights image regions that potentially influence the network's output. Some of these are true influences, but some are spurious. We then apply a causal filtering step to determine which input regions actually influence the output. This produces more succinct visual explanations and more accurately exposes the network's behavior. We demonstrate the effectiveness of our model on three datasets totaling 16 hours of driving. We first show that training with attention does not degrade the performance of the end-to-end network. Then we show that the network causally cues on a variety of features that are used by humans while driving.

As machine learning algorithms are increasingly applied to high impact yet high risk tasks, such as medical diagnosis or autonomous driving, it is critical that researchers can explain how such algorithms arrived at their predictions. In recent years, a number of image saliency methods have been developed to summarize where highly complex neural networks "look" in an image for evidence for their predictions. However, these techniques are limited by their heuristic nature and architectural constraints. In this paper, we make two main contributions: First, we propose a general framework for learning different kinds of explanations for any black box algorithm. Second, we specialise the framework to find the part of an image most responsible for a classifier decision. Unlike previous works, our method is model-agnostic and testable because it is grounded in explicit and interpretable image perturbations.

In this paper, we present a simple and modularized neural network architecture, named interleaved group convolutional neural networks (IGCNets). The main point lies in a novel building block, a pair of two successive interleaved group convolutions: primary group convolution and secondary group convolution. The two group convolutions are complementary: (i) the convolution on each partition in primary group convolution is a spatial convolution, while on each partition in secondary group convolution, the convolution is a point-wise convolution; (ii) the channels in the same secondary partition come from different primary partitions. We discuss one representative advantage: Wider than a regular convolution with the number of parameters and the computation complexity preserved. We also show that regular convolutions, group convolution with summation fusion, and the Xception block are special cases of interleaved group convolutions. Empirical results over standard benchmarks, CIFAR-10, CIFAR-100, SVHN and ImageNet demonstrate that our networks are more efficient in using parameters and computation complexity with similar or higher accuracy.

Feature selection is playing an increasingly significant role with respect to many computer vision applications spanning from object recognition to visual object tracking. However, most of the recent solutions in feature selection are not robust across different and heterogeneous set of data. In this paper, we address this issue proposing a robust probabilistic latent graph-based feature selection algorithm that performs the ranking step while considering all the possible subsets of features, as paths on a graph, bypassing the combinatorial problem analytically. An appealing characteristic of the approach is that it aims to discover an abstraction behind low-level sensory data, that is, relevancy. Relevancy is modelled as a latent variable in a PLSA-inspired generative process that allows the investigation of the importance of a feature when injected into an arbitrary set of cues. The proposed method has been tested on ten diverse benchmarks, and compared against eleven state of the art feature selection methods. Results show that the proposed approach attains the highest performance levels across many different scenarios and difficulties, thereby confirming its strong robustness while setting a new state of the art in feature selection domain.

Existing methods for visual reasoning attempt to directly map inputs to outputs using black-box architectures without explicitly modeling the underlying reasoning processes. As a result, these black-box models often learn to exploit biases in the data rather than learning to perform visual reasoning. Inspired by module networks, this paper proposes a model for visual reasoning that consists of a program generator that constructs an explicit representation of the reasoning process to be performed, and an execution engine that executes the resulting program to produce an answer. Both the program generator and the execution engine are implemented by neural networks, and are trained using a combination of backpropagation and REINFORCE. Using the CLEVR benchmark for visual reasoning, we show that our model significantly outperforms strong baselines and generalizes better in a variety of settings.

Infant recognition has received increasing attention in recent years in many applications, such as tracking child vaccination and identifying missing children. Due to the lack of efficient identification methods for infants and newborns, the current methods of infant recognition rely on identification of parents or certificates of identity. While biometric recognition technologies (e.g., face and fingerprint recognition) have been widely deployed in many applications for recognizing adults and teenagers, no such recognition systems yet exist for infants or newborns. One of the major problems is that the biometric traits of infants and newborns are either not permanent (e.g., face) or difficult to capture (e.g., fingerprint) due to lack of appropriate sensors. In this paper, we investigate the feasibility of infant recognition by their footprint using a 500 ppi commodity friction ridge sensor. We collected an infant footprint dataset in three sessions, consisting of 60 subjects, with age range from 1 to 9 months. We proposed a new minutia descriptor based on deep convolutional neural network for measuring minutiae similarity. The descriptor is compact and highly discriminative. We conducted verification experiments for both single enrolled template and fusion of multiple enrolled templates, and show the impact of age and time gap on matching performance. Comparison experiments with state of the art algorithm show the advantage of the proposed minutia descriptor.

Despite their success for object detection, convolutional neural networks are ill-equipped for incremental learning, i.e., adapting the original model trained on a set of classes to additionally detect objects of new classes, in the absence of the initial training data. They suffer from "catastrophic forgetting" - an abrupt degradation of performance on the original set of classes, when the training objective is adapted to the new classes. We present a method to address this issue, and learn object detectors incrementally, when neither the original training data nor annotations for the original classes in the new training set are available. The core of our proposed solution is a loss function to balance the interplay between predictions on the new classes and a new distillation loss which minimizes the discrepancy between responses for old classes from the original and the updated networks. This incremental learning can be performed multiple times, for a new set of classes in each step, with a moderate drop in performance compared to the baseline network trained on the ensemble of data. We present object detection results on the PASCAL VOC 2007 and COCO datasets, along with a detailed empirical analysis of the approach.

Convolutional neural networks (CNNs) provide the current state of the art in visual object classification, but they are far less accurate when classifying partially occluded objects. A straightforward way to improve classification under occlusion conditions is to train the classifier using partially occluded object examples. However, training the network on many combinations of object instances and occlusions may be computationally expensive. This work proposes an alternative approach to increasing the robustness of CNNs to occlusion. We start by studying the effect of partial occlusions on the trained CNN and show, empirically, that training on partially occluded examples reduces the spatial support of the filters. Building upon this finding, we argue that smaller filter support is beneficial for occlusion robustness. We propose a training process that uses a special regularization term that acts to shrink the spatial support of the filters. We consider three possible regularization terms that are based on second central moments, group sparsity, and mutually reweighted L1, respectively. When trained on normal (unoccluded) examples, the resulting classifier is highly robust to occlusions. For large training sets and limited training time, the proposed classifier is even more accurate than standard classifiers trained on occluded object examples.